Conventional packet-based synchronization methods such as NTP or PTP (IEEE1588) while insensitive to static delay are sensitive to packet delay variation. Packet delay variation degrades the quality of any clock signal recovered from a packet stream within the network. Conventional approaches to the mitigation of this effect require that every node in the network is protocol aware, and thus in some cases actively modify the synchronization packet stream, e.g. the PTP transparent clocks.
Segmentation of the synchronization network, e.g. IEEE1588 boundary clocks for providing residency time correction as in IEEE1588 transparent clocks can require that every device or node in the synchronization path forms an active participant in the timing protocol. An alternative conventional approach is to provide that every device in the path for the synchronization stream does have a fixed delay. This is typically but not always achieved by minimizing the delay in each device using traffic management techniques. Further, some conventional dynamic buffer allocation techniques have been proposed for specific applications relating to voice transport or circuit emulation services. However, these conventional dynamic buffer allocation techniques depend on the rate of transmission of packets of interest being constant and do not employ timestamps.
Accordingly, there is a need for a method and apparatus for mitigation of packet delay variation in a packet-based network which overcome the above-mentioned drawbacks and which can minimize a packet delay variation within the network.